Coherent communication systems are being designed in the long wavelength regime 1.2 .mu.m to 1.6 .mu.m using Group III-V semiconductor lasers to take advantage of optical fiber loss minima. In these applications, it is necessary to maintain reasonably close tolerances on wavelength (frequency) stability of the optical sources. Unfortunately, semiconductor lasers exhibit sufficiently large frequency drifts due to junction temperature and injection current variations to obstruct their use as transmitter sources, receiver local oscillators, or system references or reference clocks in such systems without the addition of complicated compensation or tracking control loops.
It is understood that because of their well understood mathematical relationship, the terms "wavelength" and "frequency" are used interchangeably herein without loss of generality.
Frequency stabilization techniques for semiconductor lasers have been widely discussed in the technical literature since the early part of the present decade. While the scope of techniques discussed is quite broad to include the use of Fabry-Perot interferometers and atomic or molecular transition lines as references, most reports are limited to the short wavelength region involving AlGaAs lasers which operate around 0.8 .mu.m. See, for example, C. J. Nielsen et al., J. Opt. Commun., Vol. 4, pp. 122-5 (1983) dealing with the use of Fabry-Perot interferometers; S. Yamaguchi et al., IEEE J. Quantum Electron. Vol. QE-19, pp. 1514-9 (1983) on the use of atomic transition lines; and H. Tsuchida et al., Jpn. J. Appl. Phys., Vol. 21, pp. L1-L3 (1982) on the use of molecular transition lines.
Fabry-Perot interferometric techniques are vulnerable to long term drifts caused by fluctuations of the resonant cavity. Longer term stability is afforded by using atomic or molecular transition lines. Of the latter two approaches, atomic transition lines are preferred over molecular transition lines. Atomic spectra offer relatively few, widely separated and, therefore, easily identifiable strong lines. On the other hand, molecular transition line spectra are complex and weak which, in turn, results in the need for very long absorption cells as a reference.
Only a few articles have reported experiments for laser frequency stabilization on longer wavelength semiconductor lasers operating above 1.2 .mu.m. Such sources are usually based on Group III-V materials such as InGaAsP/InP. The reported experiments solely employ molecular transition lines as references. In one experiment, absorption lines of the ammonia (NH.sub.3) molecule are used to frequency stabilize an InGaAsP distributed feedback laser. See, T. Yanagawa et al., Appl. Phys. Lett., Vol. 47, pp. 1036-8 (1985). Another experiment employed first overtone vibration-rotation lines of hydrogen fluoride (HF) molecules to frequency lock an InGaAsP laser. See, S. Yamaguchi et al., Appl. Phys. Lett., Vol. 41, pp. 1034-6 (1982). One other experiment utilized absorption lines of water vapor (H.sub.2 O) molecules and ammonia molecules in spectral measurements for pollutant gas monitoring. See M. Ohtsu et al., Jpn. J. Appl. Phys., Vol. 22, pp. 1553-7 (1983).
It is noteworthy that experiments reported in the long wavelength region are restricted to using only molecular transition lines. In part or in whole, this is apparently due to the fact that there has been much reported difficulty, and there have been no reported successes, in finding useful atomic transitions in this wavelength region emanating from the ground state. But, a careful review of the literature shows that one need not be restricted to only those frequency stabilization techniques which employ atomic transitions from the ground state. It is well known to use optogalvanic signals corresponding to transitions from excited atomic states for stabilizing short wavelength AlGaAs semiconductor lasers. See, for example, S. Yamaguchi et al., IEEE J. Quantum Electron., Vol. QE-19, 1514-9 (1983) describing the optogalvanic effect of krypton.
Unfortunately, extension of the optogalvanic effect to frequency stabilization techniques for longer wavelength semiconductor lasers was cast in serious doubt in 1982 when the recognized experts in the field stated that output power from longer wavelength semiconductor lasers (InGaAsP/InP) is insufficient to produce an impedance change in the discharge tube. Since the optogalvanic effect concerns large impedance changes of the gas discharges by optical (laser) irradiation at wavelengths corresponding to non-ionizing transitions of species present in the discharge, the experts deduced that the optogalvanic signals would be difficult, at best, to detect for the longer wavelength lasers. See, S. Yamaguchi et al., Appl. Phys. Lett., Vol. 41, pp. 1034 (1982). It is reasonable to surmise that the failure to have published works over the last score on the extension of the optogalvanic signal techniques for frequency stabilization developed and improved by Yamaguchi et al. at short wavelengths is, in large part, due to the chilling effect on other researchers of the experts' own published statement which knowingly predicts failure of the technique at longer wavelengths.